Cells of Chironomus salivary glands and Malpighian tubules have junctions of the "septate" kind. This is the only kind of junction discerned which is large enough to effect the existing degree of intercellular communication. The electron microscopic observations of the "septate" junction conform to a honeycomb structure, with 80-A-thick electron-opaque walls and 90-A-wide transparent cores, connecting the cellular surface membranes. A projection pattern of light and dark bands (the " septa ") with a 150-A periodicity results when the electron beam is directed normal to any set of honeycomb walls. Treatment of the salivary gland cells with media, which interrupt cellular communication (without noticeable alteration of cellular adhesion) by reducing junctional membrane permeability or perijunctional insulation, produces no alterations in the junctional structure discernible in electron micrographs of glutaraldehyde-fixed cell material.

Intercellular communication was examined with intracellular electrical techniques in primary and transplanted rat liver cancers. Normal liver cells communicate rather freely with each other through permeable junctional membranes. Cancer liver cells show no communication at all; their surface membrane is a strong barrier to diffusion all around the cell. Cancer cells induce alterations in membrane permeability in normal liver cells; communication among the latter is markedly reduced when cancer cells grow near them.

Intercellular communication was examined in regenerating rat liver and urodele skin, two tissues of fast but normal growth. In both, cellular communication is in general as good as in their respective normal intact state. This stands in striking contrast to the lack of cellular communication in tissues with cancerous growth. Upon wounding of the urodele skin, the normally permeable junctional membranes of cells near the wound border seal themselves off, thereby insulating the interiors of the communicated cell systems from the exterior. When the cells of two opposing borders make mechanical contact in the course of wound closure, communication between them ensues within 30 min. Within this period all cell movement also ceases ("contact inhibition"). The possible implications of these findings in the control of tissue growth are discussed.

The fine structures of nuclear envelopes known to have different permeability properties were compared. Membranes of salivary gland cell nuclei of Drosophila (third instar) and Chironomus (prepupae), which are strong barriers to ion diffusion, and membranes of oocyte nuclei (germinal vesicle) of Xenopus and Triturus , which are much more ion-permeable, show no essential difference in size, frequency, and distribution of their membrane gaps ("pores") which could account for the marked disparities in membrane permeability. The gaps are occupied by diffuse electron-opaque material with occasional central regions of strong opacity. This material may possibly account for the high diffusion resistance of Drosophila and Chironomus nuclear envelopes, where the resistance is far too great to allow free diffusion through the gaps. But material of this kind is also present in the more permeable nuclear envelopes of Xenopus and Triturus oocytes, and there are no convincing structural differences discernible with the techniques employed.

Membrane permeability of an epithelial cell junction ( Drosophila salivary gland) was examined with intracellular microelectrodes and with fluorescent tracers. In contrast to the non-junctional cell membrane surface, which has a low permeability to ions (10 -4 mho/cm 2 ), the junctional membrane surface is highly permeable. In fact, it introduces no substantial restriction to ion flow beyond that in the cytoplasm; the resistance through a chain of cells (150 Ω cm) is only slightly greater than in extruded cytoplasm (100 Ω cm). The diffusion resistance along the intercellular space to the exterior, on the other hand, is very high. Here, there exists an ion barrier of, at least, 10 4 Ω cm 2 . As a result, small ions and fluorescein move rather freely from one cell to the next, but do not leak appreciably through the intercellular space to the exterior. The organ here, rather than the single cell, appears to be the unit of ion environment. The possible underlying structural aspects are discussed.

The surface structure of a gland epithelium ( Drosophila salivary gland), particularly that at the junction between cells, was examined under the electron microscope. The junctional surface, which in the preceding paper was shown to be highly permeable to ions, has the following structural characteristics. About two-thirds of it are profusely infolded; the surface membranes of adjoining cells interdigitate and present desmosomes. The width of the intercellular space varies considerably. The remainder of the junctional surface, the third that abuts on the lumen, is rather straight. Here, the cell membranes are aligned parallel at a distance of 150 A, and interconnected at regular intervals of 100 A. The connecting material has a high electron opacity, and is about as thick as the cell membranes, but, unlike the latter, has no resolvable unit membrane structure. The surface at the cell base, which in the preceding paper was shown to be rather impermeable, is infolded and resembles the infolded junctional region. The luminal surface exhibits microvilli. Critical surface dimensions are given, and the implications of surface structure in intercellular permeability are discussed.